Correcting What You Cannot See: Credit Assignment for Perception Distillation in Multimodal Reasoners

• URL: http://arxiv.org/abs/2607.28336v2
• Date: Sat, 01 Aug 2026 07:38:13 GMT
• Title: Correcting What You Cannot See: Credit Assignment for Perception Distillation in Multimodal Reasoners
• Abstract summary: We introduce textbfPerception-Correction Distillation (PCD), a label-free method that identifies correctable perception failures.<n>We motivate this rule through Bayesian evidence combination and show that multiplication is the unique normalized bilinear gate that vanishes when either witness is absent.
• Score: 34.39191083253229
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: On-policy distillation provides dense supervision for multimodal reasoners, but its trajectory-level reward cannot determine whether a failed answer arose from perception or subsequent reasoning. Perception Success Rate (PSR), estimated from multiple reasonings sharing one perception, remains ambiguous because low success conflates perceptual insufficiency with reasoning difficulty. We introduce \textbf{Perception-Correction Distillation (PCD)}, a label-free method that identifies correctable perception failures using downstream failure and teacher--student disagreement as complementary witnesses. Their product, , forms a soft AND gate that strengthens distillation only when both witnesses are present. We motivate this rule through Bayesian evidence combination and show that multiplication is the unique normalized bilinear gate that vanishes when either witness is absent. PCD uses separated perception--reasoning rollouts and mean-preserving weights, leaving the reasoning objective unchanged. Across eight benchmarks, PCD improves the 8B 2B macro average from 44.50 with OPD to 47.28 and the 32B 8B result from 56.94 to 61.22. In matched 2B ablations, removing PCD and separated rollout reduces held-out average by 2.22 and 0.88 points, respectively. Effective multimodal distillation therefore depends not only on what the teacher predicts, but also on identifying when perception is the appropriate target of correction.

Related papers

• Reliable Benchmarking of Artifact Detection in Computational Pathology: A Reproducibility and Uncertainty Analysis [1.8864137667201766]
  Methods: The protocol quantifies four sources of variability - test-set sampling, trainingity, partition composition, and undocumented preprocessing.<n>We apply it to an independent reconstruction of a published diffusion-based artifact detector, evaluated on the original 24-slide partition and against a supervised baseline.<n>Results: The method's central mechanism reproduces: the auxiliary contrastive term improves pooled F1 from 0.673 to 0.688 and replicates under a second seed.
  arXiv  Detail & Related papers  (2026-08-31T14:07:15Z)
• The Distributional View of Knowledge Distillation [13.841028090842869]
  Token-level knowledge distillation matches two conditional distributions per position.<n>We develop a distributional view in which the teacher is represented not by a single softened output but by a family of multi-temperature views.
  arXiv  Detail & Related papers  (2026-08-15T13:03:31Z)
• When Outputs Disperse, Does Epistemic Revision Follow? A Black-Box Diagnostic for Machine Collectives [0.0]
  We operationalize dispersion-revision coupling in collective intelligence research.<n>We show that conditional dissent improves false-premise recovery by +17.7 points.<n>We recommend reporting per-intervention stance shift and premise-preservation rate alongside accuracy.
  arXiv  Detail & Related papers  (2026-08-04T14:19:59Z)
• Outcome-Confounded Local Supervision in On-Policy Distillation [0.5937989874256571]
  On-policy distillation trains a student on its own trajectories while a teacher supplies dense token-level likelihoods at student-visited prefixes.<n>We show that both readings are confounded by the outcome of the completed trajectory.<n>We introduce an outcome-resolved diagnostic that crosses pointwise teacher-student divergence with final-answer correctness.
  arXiv  Detail & Related papers  (2026-07-26T16:03:33Z)
• Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning [73.40085256022111]
  We propose Perceive-to-Reason (P2R), a unified framework that formulates fine-grained visual reasoning as a two-stage process.<n>P2R localizes question-relevant evidence as a Perceiver, and then answers the question as a Reasoner based on the annotated image and cropped regions.<n>PRA-GRPO is a role-aware reinforcement learning strategy that alternates between perception-focused and reasoning-focused updates.
  arXiv  Detail & Related papers  (2026-07-01T17:24:26Z)
• CRAFT: Counterfactual Credit Assignment from Free Sibling Rollouts for Self-Distilled Agentic Reinforcement Learning [4.663685189987781]
  Self-distilled agentic reinforcement learning augments trajectory-level reward with a token-level distillation loss.<n>The prevailing recipe gates this loss by a single scalar, the teacher-student log-probability gap.<n>We introduce CRAFT, a threepillar credit-assignment scheme that addresses both limitations.
  arXiv  Detail & Related papers  (2026-06-28T16:11:47Z)
• CoDMD: Copula-aware Distribution Matching Distillation for Fast Video Generation [50.353919095724315]
  Few-step distillation for video diffusion models has attracted significant attention, driven by the urgent demand for efficient deployment in real-world scenarios.<n>We propose Copula-aware DMD (CoDMD), a lightweight relational regularizer that reuses score estimates already produced by the frozen teacher and the online fake model to construct pairwise relation matrices.<n>On the Wan-2.1-T2V model series at 1.3B & 14B scales, CoDMD distills 50-step teachers into 4-step students, achieving an approximate 25$times$ speed-up while attaining VBench scores of 84.46 & 84.87,
  arXiv  Detail & Related papers  (2026-06-20T10:33:14Z)
• Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning [55.264863369127774]
  Current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.<n>We propose Hidden-Align, an auxiliary loss function that aligns the last-layer hidden states of correct rollouts at the anchor token during RL training, with zero overhead in both training and inference.<n>On eight mathematical reasoning benchmarks, Hidden-Align improves average pass@1 over the DAPO baseline by 3.8, 6.2, and 5.4 percentage points on Qwen3-1.7B, 4B, and 14B respectively, with consistent pass@k gains across all three scales.
  arXiv  Detail & Related papers  (2026-06-02T06:51:15Z)
• MAIGO: Mitigating Lost-in-Conversation with History-Cleaned On-Policy Self-Distillation [51.32912774464992]
  Large language models often solve tasks from a fully specified prompt but degrade when the same requirements unfold over multiple turns.<n>We propose MAIGO, an on-policy self-distillation method that reduces this contamination using history-cleaned references from the model's own policy.<n>For middle turns, MAIGO removes prior assistant replies while preserving the user-visible sharded prefix; for answer turns, it distills from paired full-view references conditioned on the completed user-side dialogue.
  arXiv  Detail & Related papers  (2026-05-26T15:38:46Z)
• Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy [0.7212939068975618]
  We find that pretrained base models exhibit the same substitution pattern as their Instruct variants, averaging higher yield than Instruct.<n> patching above this window restores 96% of the clean-to-pressured P(correct) gap.<n>Two converging activation-space interventions show that pressure suppresses clean-reasoning features rather than activating a new sycophancy circuit.
  arXiv  Detail & Related papers  (2026-05-13T04:45:08Z)
• Correct, Concise and Complete: Multi-stage Training For Adaptive Reasoning [11.179446105672461]
  We propose a multi-stage efficient reasoning method that combines supervised fine-tuning and reinforcement learning.<n>Our approach reduces response length by an average of 28% for 8B models and 40% for 32B models.<n>It achieves a superior trade-off compared to more complex state-of-the-art efficient reasoning methods.
  arXiv  Detail & Related papers  (2026-01-06T12:31:51Z)
• CARE What Fails: Contrastive Anchored-REflection for Verifiable Multimodal [84.71254539482369]
  Group-relative reinforcement learning with verifiable rewards (RLVR) often wastes the most informative data it already has the failures.<n>We present CARE, a failure-centric post-training framework for multimodal reasoning that turns errors into supervision.<n> CARE improves accuracy and training smoothness while explicitly increasing the share of learning signal that comes from failures.
  arXiv  Detail & Related papers  (2025-12-22T16:34:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.